Transcriptomic and Quantitative Proteomic Profiling Reveals Signaling Pathways Critical for Pancreatic Islet Maturation

Abstract

Pancreatic β-cell dysfunction and reduced insulin secretion play a key role in the pathogenesis of diabetes. Fetal and neonatal islets are functionally immature and have blunted glucose responsiveness and decreased insulin secretion in response to stimuli and are far more proliferative. However, the mechanisms underlying functional immaturity are not well understood. Pancreatic islets are composed of a mixture of different cell types, and the microenvironment of islets and interactions between these cell types are critical for β-cell development and maturation. RNA sequencing and quantitative proteomic data from intact islets isolated from fetal (embryonic day 19) and 2-week-old Sprague-Dawley rats were integrated to compare their gene and protein expression profiles. Ingenuity Pathway Analysis (IPA) was also applied to elucidate pathways and upstream regulators modulating functional maturation of islets. By integrating transcriptome and proteomic data, 917 differentially expressed genes/proteins were identified with a false discovery rate of less than 0.05. A total of 411 and 506 of them were upregulated and downregulated in the 2-week-old islets, respectively. IPA revealed novel critical pathways associated with functional maturation of islets, such as AMPK (adenosine monophosphate-activated protein kinase) and aryl hydrocarbon receptor signaling, as well as the importance of lipid homeostasis/signaling and neuronal function. Furthermore, we also identified many proteins enriched either in fetal or 2-week-old islets related to extracellular matrix and cell communication, suggesting that these pathways play critical roles in islet maturation. Our present study identified novel pathways for mature islet function in addition to confirming previously reported mechanisms, and provided new mechanistic insights for future research on diabetes prevention and treatment.

Diabetes is a major health care problem and is a leading cause of adult blindness, kidney failure, heart disease, peripheral neuropathy, and premature mortality (1). The prevalence of diabetes worldwide is increasing rapidly, and as of 2019 approximately 463 million adults had diabetes (2). β-Cell dysfunction and reduced insulin secretion play a key role in the pathogenesis of diabetes; however, the underlying mechanisms causing β-cell failure are complex and remain to be fully elucidated. Thus, it is critically important to better understand the molecular mediators regulating β-cell development and function to better prevent and treat diabetes.

When the blood glucose concentration increases after a meal, glucose uptake and metabolism by β cells increases, which initiates insulin secretion. Unlike adult β cells, fetal and neonatal β cells have blunted glucose responsiveness and decreased insulin secretion in response to secretagogues (3-8). However, glucose-stimulated insulin secretion (GSIS) rapidly changes to a more adult pattern after birth. The mechanisms underlying the functional immaturity of fetal and neonatal islets are not well understood. In part, blunted mitochondrial function, including differences in islet respiration, nucleotide turnover, glucose metabolism, dysregulation of adenosine 5′-triphosphate–sensitive potassium channels, and a low expression and activity of nicotinamide adenine dinucleotide (NADH) shuttle enzymes and mitochondrial adenine-nucleotide translocator, all contribute to the immaturity of fetal and neonatal islets (8-12). Results of a study comparing the transcriptome profiles of neonatal and adult laser-captured β cells using microarrays suggested that low expression of key metabolic enzymes, in particular, mitochondrial shuttles, pyruvate carboxylase, and carnitine palmitoyl transferase 2, contributes to the dampened glucose-responsiveness of neonatal β cells (13).

Recently it was shown that overexpression of c-Myc, a cell cycle regulator, induces β cells to enter the cell cycle, which in turn promotes a functionally immature phenotype. These results suggest that proliferation and maturity are negatively correlated and may be mutually exclusive states in β cells (14).

Although insulin-producing β cells are the major cell type and represent 60% to 80% of cells in the islets of rodents, other hormone-producing cells are also present in islets, such as α cells (glucagon), δ cells (somatostatin), ε cells (ghrelin), and PP-cells (pancreatic polypeptides) (15, 16). The islet functions as a mini-organ and is highly vascularized, innervated by sympathetic, parasympathetic, and sensory neurons, and contains residential immune cells (17-20). The microenvironment of islets and the interactions among different cell types are critical for β-cell development and maturation, as well as maintaining glucose homeostasis.

At 2 weeks of age, insulin secretion has not quite achieved adult levels but is markedly higher than fetal levels (7). We hypothesized that integrating RNA sequencing (RNA-seq) and quantitative proteomic data from intact islets isolated from fetal (embryonic day 19 [e19]) and 2-week-old (2-wk) rats would identify novel molecular mechanisms controlling islet functional development during the transition from fetal to postnatal life. In addition to confirming previously reported genes and pathways that were altered during β-cell development, we identified many proteins enriched in islets at different ages and signaling pathways and upstream regulators critical for β-cell/islet maturation and function.

Materials and Methods

Islet isolation

The animals and procedures used in this study were approved by the Animal Care Committee of The Children’s Hospital of Philadelphia and the Perelman School of Medicine at the University of Pennsylvania. Pancreata were excised from Sprague Dawley rats (Charles River) under ketamine and xylazine anesthesia at e19 and age 2 weeks followed by islet isolation. Pancreata from a litter were pooled for each e19 sample; approximately 2 to 4 pancreata per litter from both sexes were pooled for each 2-week sample. Pancreatic islets were isolated as previously described (21). Briefly, pancreata were digested with Collagenase P (Roche) in Hank’s balanced salt solution (HBSS) supplemented with 4 mM NaCO₃ and 1% bovine serum albumin for 10 to 15 minutes at 37°C. Digested tissues were then washed in cold supplemented HBSS solution without collagenase. Islets were isolated by histopaque gradient centrifugation. Purity of the islet preparations were confirmed as previously described (21) and assessed for presence of insulin and elastase 3B using enzyme-linked immunosorbent assays.

Total RNA isolation and RNA sequencing library preparation

Total RNA was extracted from freshly isolated islets (n = 3) using TRIzol Reagent (Invitrogen), followed by Qiagen RNeasy Mini kit following manufacturer’s instructions. RNA samples with integrity numbers between 7 and 9.5 were used for RNA-seq. Libraries were generated using the Illumina TruSeq Stranded Total RNA LT Sample Prep Kit with Ribo-Zero Gold.

RNA sequencing and gene expression analysis

RNA-seq and gene expression were analyzed as previously described (22). Briefly, RNA-seq libraries were single-end sequenced to 100 bp on an Illumina hiSeq2000. All RNA-seq data were mapped using the TopHat package for the rat genome (rn5). Differential analysis was performed using EdgeR. Differentially regulated genes were identified using a false discovery rate (FDR, q value) cutoff of 0.05. Clustering heatmaps were constructed using z score that was scaled across samples for each gene. Functional analysis using QIAGEN’s Ingenuity Pathway Analysis (IPA) was performed on genes with a fold change greater than or equal to 1.5, cpm 2 or greater, and a q value of less than .05. The data were deposited in NCBI’s (National Center for Biotechnology Information’s) Gene Expression Omnibus and are accessible through GEO Series accession number GSE153604.

Sample preparation for proteomics

Frozen islet samples (n = 4) were sent to the Proteomics Core Facility at the Children’s Hospital of Philadelphia for protein hydrolysis, followed by peptide separation, and analyzed by liquid chromatography with tandem mass spectrometry (MS) on a QExactive HF mass spectrometer (Thermo Fisher Scientific) coupled with an Ultimate 3000. The label-free approach was chosen for its adaptability to include new samples when needed, as well as to avoid the possible errors while labeling techniques were applied.

Protein sequence database search and proteomics analysis

MS/MS raw files were searched against a human protein sequence database including isoforms from the Uniprot Knowledgebase (taxonomy:10090 AND keyword: “Complete proteome [KW-0181]”) using MaxQuant (23) version 1.6.1.0 with the following parameters: fixed modifications, carbamidomethyl (C); decoy mode, revert; MS/MS tolerance Fourier-transform mass spectrometry 20 ppm; FDR for both peptides and proteins of 0.01; minimum peptide length of 7; modifications included in protein quantification, acetyl (protein N-term), oxidation (M); peptides used for protein quantification, razor and unique. iBAQ values were used for protein quantification.

Perseus (version 1.6.1.1) was used for proteomic data processing and statistical analysis. Protein groups containing matches to decoy database or contaminants were discarded. The data were log₂-transformed and normalized by subtracting the median for each sample. The t test was employed to identify differentially expressed proteins. The Benjamini-Hochberg approach was applied to obtain FDR.

Results

Transcriptome profiles differ between fetal and 2-week-old islets

The principal component analysis (PCA plot, Fig. 1A) and hierarchical clustering analysis of differentially expressed genes (Fig. 1B) showed a clear separation between fetal and 2-wk islets. In total, 5896 transcripts were differentially expressed in 2-wk compared to fetal islets (Supplemental Table S1) (24). As shown in the volcano plot (Fig. 1C), 2763 and 3133 transcripts were significantly (q < 0.05) increased and decreased in 2-wk islets, respectively. We have demonstrated in our previous RNA-seq study (22) that the changes for differentially expressed genes in the RNA-seq data set were consistent with the changes determined via quantitative polymerase chain reaction. To identify pathways regulating islet development and maturation, IPA was used for functional annotation of differentially expressed genes. More than 180 canonical pathways were altered in 2-wk islets. Many of them are well-known pathways critical for β-cell development, such as cell cycle regulation, translation and replication, and pathways regulating mitochondrial function, tricarboxylic acid cycle, glycolysis, and glucose sensing (Supplemental Table S2) (24). IPA also revealed pathways that may play important roles for islet development and maturation, including pathways regulating lipid metabolism and signaling, neuronal function, immune function, and vascularization.

Figure 1.

Transcriptome of 2-week islets compared to fetal islets. A, Principal component analysis of fetal and 2-week islets showing a clear difference in transcriptome profiles. B, Heat map of differentially expressed genes in 2-week islets compared with fetal islets. Each row in the heat map corresponds to data from a single gene, whereas columns correspond to individual samples. The branching dendrogram corresponds to the relationships among samples, as determined by clustering using 5896 differentially expressed genes. Increases and decreases of gene expression are shown on a continuum from red to blue, respectively. C, Volcano plot visualizing the changes in differentially expressed genes. Genes whose expression was significantly increased at 2 weeks are in the top section, and genes significantly decreased are in the bottom section.

Proteomic profiles differ between fetal and 2-week-old islets

We identified 8014 proteins in fetal islets and 7930 proteins in 2-week-old rats. Principal component analysis (Fig. 2A) showed a clear difference in proteomic profiles between fetal and 2-wk islets. Unsupervised hierarchical cluster analysis (Fig. 2B) showed consistent patterns of protein expression across all samples, without an overall change of protein abundance in 2-wk islets. Using the inclusion criteria of q less than 0.05 to analyze differentially expressed proteins, we identified 2149 proteins that were differentially expressed in 2-wk islets compared with fetal islets (Supplemental Table S3) (24). Hierarchical clustering of differentially expressed proteins readily separated the 2 groups (Fig. 2C). Among them, 994 and 1155 proteins were increased and decreased in 2-wk islets compared to fetal islets, respectively (Fig. 2D). IPA revealed that more than 140 canonical pathways were significantly different. Nearly 80 of them were the same pathways identified in our RNA-seq analysis, including pathways regulating cell cycle, replication, mitochondrial function, glucose sensing, insulin signaling, sex hormone signaling, cholesterol biosynthesis, and neuronal function (Table 1).

Figure 2.

Proteomics of 2-week islets compared to fetal islets. A, Principal component analysis (PCA) of fetal and 2-week islets showing a clear difference in proteomic profiles. B, An unsupervised clustering heat map showing no change in protein abundance between 2-week and fetal islets. Each row in the heat map corresponds to data from a single protein. The branching dendrogram at the top corresponds to the relationships among samples, as determined by unsupervised clustering. Increases and decreases in protein abundance are shown on a continuum from red to green, respectively. C, Heat map of differentially expressed proteins showing significant differences in abundance between 2-week and fetal islets. Each row in the heat map corresponds to data from a single protein, whereas columns correspond to individual samples. The branching dendrogram corresponds to the relationships among samples, as determined by clustering using identified 2149 differentially expressed proteins. Increases and decreases in protein abundance are shown on a continuum from red to green, respectively. D, Volcano plot visualizing the changes in differentially expressed proteins. Proteins significantly more abundant in 2-week islets are in the upper right section, and proteins significantly more abundant in fetal islets are in the upper left section. The difference (E19 – 2 week) at the x-axis represents log fold change in 2-week islets compared with fetal islets.

Table 1.

Top canonical pathways overlapping in the transcriptome and proteome of islets

Ingenuity canonical pathways	RNA-Seq		Proteomics
	P	z score	P	z score
AMPK signaling	1.26E-03	2.71	8.13E-06	2.12
Androgen signaling	1.29E-04	1.89	1.07E-02	2.65
Aryl hydrocarbon receptor signaling	1.41E-04	–1.30	2.24E-06	–1.89
ATM signaling	2.57E-04	–1.46	4.57E-05	–1.61
Cardiac hypertrophy signaling	2.45E-02	4.57	4.17E-02	0.78
Cell cycle control of chromosomal replication	5.75E-06	–5.39	8.71E-10	–
CREB signaling in neurons	8.13E-03	3.04	3.39E-02	1.89
Cyclins and cell cycle regulation	4.90E-06	–3.41	1.26E-02	–2.11
Dopamine receptor signaling	3.16E-02	1.41	2.69E-06	1.00
Estrogen receptor signaling	2.40E-03	1.11	1.86E-05	–
Gluconeogenesis I	2.95E-02	–0.90	1.48E-03	2.83
Glycolysis I	1.07E-02	–1.15	6.61E-03	2.65
Insulin receptor signaling	1.58E-02	–0.93	6.03E-04	–1.63
Mitotic roles of Polo-like kinase	2.09E-10	–3.00	1.86E-03	–2.65
NRF2-mediated oxidative stress response	3.98E-04	–1.00	3.31E-02	–1.90
PI3K/AKT signaling	3.09E-02	–0.52	1.29E-02	–1.89
PPARα/RXRα activation	1.41E-02	–0.90	1.15E-04	1.22
Protein kinase A signaling	7.59E-04	2.69	4.57E-03	2.47
Sumoylation pathway	2.09E-03	1.30	3.24E-05	1.79
Superpathway of cholesterol biosynthesis	5.13E-09	–4.15	5.25E-04	–0.33
β-Adrenergic signaling	6.92E-04	3.09	7.41E-04	2.06

Abbreviations: AKT, protein kinase B; AMPK, adenosine monophosphate–activated protein kinase; ATM, ataxia telangiectasia mutated; CREB, 3′,5′-cyclic AMP‐response element‐binding protein; PI3K, phosphatidylinositol-3-kinase; PPAR, peroxisome proliferator-activated receptor; RNA-Seq, RNA sequencing; RXR, retinoid X receptor.

Pathways critical for islet development and mature function

Integrating RNA-seq and proteomic datasets identified 917 differentially expressed genes (q < 0.05, fold change ≥1.5, cpm ≥ 2) that correlated with significant changes (q < 0.05) in the same direction in protein abundance in 2-wk islets compared to fetal islets (Supplemental Table S4) (24). A total of 411 and 506 genes/proteins were upregulated and downregulated in 2-wk islets, respectively. Differentially expressed genes/proteins whose changes in protein abundance did not correlate with changes in RNA levels are listed in Supplemental Table S5 (24). IPA of these 917 genes/proteins revealed 71 significantly altered canonical pathways in 2-wk islets. IPA analysis identifies biological pathways that are expected to be increased or decreased given the observed gene expression changes in the data set. For each biological function a statistical quantity is computed, called the activation or inhibition z score based. As predicted by activation z score, top pathways activated in 2-wk islets included retinol biosynthesis, triacylglycerol degradation, synaptogenesis signaling, sumoylation pathway, death receptor signaling, and adenosine monophosphate–activated protein kinase (AMPK) signaling (Table 2). In contrast, the top pathways that were predicted to be inhibited at 2 weeks compared to fetal islets included pathways controlling cell cycle/replication/translation, nucleotide excision repair (NER) pathway, super-pathway of cholesterol biosynthesis, unfolded protein response, and aryl hydrocarbon receptor (AhR) signaling (see Table 2). Other interesting pathways altered in 2-wk islets included dopamine receptor signaling, adipogenesis pathway, Farnesoid X receptor (FXR)/retinoid X receptor activation, thyroid hormone receptor (TR)/retinoid X receptor (RXR) activation, and glucocorticoid receptor signaling (see Table 2).

Table 2.

Top canonical pathways identified by integrating the transcriptome and proteome

Ingenuity canonical pathways	P	Activation z score
Cell cycle control of chromosomal replication	2.51E-12	–4.24
NER pathway	1.00E-11	–4.02
EIF2 signaling	2.40E-04	–2.50
Superpathway of geranylgeranyldiphosphate biosynthesis I (via mevalonate)	5.37E-04	–2.24
Aryl hydrocarbon receptor signaling	5.62E-04	–1.15
Retinol biosynthesis	1.20E-03	2.65
Synaptogenesis signaling pathway	1.58E-03	2.13
Mevalonate pathway I	1.78E-03	–2.00
Estrogen-mediated S-phase entry	3.16E-03	–1.34
Mitotic roles of Polo-like kinase	4.37E-03	–2.00
ATM signaling	5.01E-03	–1.90
Superpathway of cholesterol biosynthesis	5.13E-03	–2.24
Sumoylation pathway	7.76E-03	1.90
Triacylglycerol degradation	1.02E-02	2.45
Glucocorticoid receptor signaling	1.51E-02	–
Adipogenesis pathway	1.74E-02	–
TR/RXR activation	1.82E-02	–
Unfolded protein response	2.29E-02	–1.34
Death receptor signaling	2.75E-02	1.41
FXR/RXR activation	2.82E-02	–
Dopamine receptor signaling	3.24E-02	–
Cyclins and cell cycle regulation	4.17E-02	–2.45
Regulation of cellular mechanics by calpain protease	4.37E-02	-0.45
AMPK signaling	4.57E-02	1.00
MODY signaling	4.79E-02	–

Abbreviations: AMPK, adenosine monophosphate–activated protein kinase; ATM, ataxia telangiectasia mutated; EIF2, eukaryotic initiation factor 2; FXR, Farnesoid X receptor; MODY, maturity onset diabetes of young; NER, nucleotide excision repair; RXR, retinoid X receptor; TR, thyroid hormone receptor.

Adenosine monophosphate–activated protein kinase signaling.

AMPK signaling was predicted to be activated in 2-wk islets (z score = 1.00). AMPK is a master metabolic regulator controlling glucose and lipid metabolism. AMPK also plays a critical role in maintaining β-cell identity. Loss of AMPK in β cells upregulates many β-cell–disallowed genes—genes that are used for housekeeping functions in other tissues but selectively repressed in islets. Examples of these disallowed genes include Slc16a1, Ldha, Mgst1, and Pdgfrα (25). Fourteen genes/proteins comprising this pathway were differentially expressed in 2-wk islets, including Ppm1e, Pfkfb2, Ins1, Prkacb, Rab3a, and Slc2a1 (Fig. 3A, Supplemental Table S6) (24).

Figure 3.

Ingenuity pathway analysis canonical pathway annotation of differentially expressed genes/proteins that are involved in the A, AMPK (adenosine monophosphate–activated protein kinase) signaling; B, Thyroid hormone receptor (TR)/retinoid X receptor (RXR) and Farnesoid X receptor (FXR)/RXR signaling; C, Aryl hydrocarbon receptor (AhR) signaling; and D, neuronal-related signaling. Red-filled and green-filled shapes mark increased and decreased expression, respectively. Oval shape represents transcription regulator; diamond represents enzyme; triangle represents phosphatase; upside down triangle represents kinase; dotted rectangle represents ion channel; trapezoid represents transporter; circle represents others.

Lipid homeostasis and signaling.

Retinoids are lipophilic molecules and metabolites of vitamin A (all-trans-retinol). They modulate a wide variety of biological processes, including cell proliferation, differentiation, and apoptosis. These retinoids function as important regulators both during embryogenesis and in the adult (26, 27). They also play an important role in regulating energy metabolism, and are required for the development of pancreatic islet and β cells in utero and maintenance of normal endocrine function in adult pancreas (28, 29). Retinol biosynthesis was predicted to be activated in 2-wk islets (z score = 2.65), and 7 genes/proteins regulating this pathway were differentially expressed, including Ddhd2, Pnliprp2, Cel, Rbp1, Pnlip, Lipe, and Lpl.

The actions of retinoids are mediated through retinoic acid receptors and RXRs (30). RXRs are common heterodimer partners for multiple nuclear receptors, such as peroxisome proliferator-activated receptors (PPARs), liver X receptors (LXRs), TR, and FXR (31). Both TR/RXR (P = 0.02) and FXR/RXR (P = 0.03) signaling were altered in 2-wk islets. A total of 16 genes/proteins comprising these pathways were differentially expressed, including Gc, Ins1, Rab3b, Trh, and Slc2a1 (Fig. 3B, Supplemental Table S7) (24).

Interestingly, many pathways regulating cholesterol and lipid homeostasis were altered in 2-wk islets, including the super pathway of cholesterol biosynthesis, triacylglycerol degradation, super pathway of geranylgeranyldiphosphate biosynthesis, adipogenesis pathway, and mevalonate pathway. Twenty-one genes/proteins comprising these pathways were differentially expressed in 2-wk islets compared with fetal islets, such as Pnliprp2, Pnlip, Lpl, Hmgcs1, Acat2, Fabp4, and Gtf2h1 (Supplemental Table S8) (24).

Aryl hydrocarbon receptor signaling.

AhR signaling was predicted to be inhibited in 2-wk islets with a z score of –1.15. AhR is a transcription factor that regulates expression of metabolic enzymes and genes critical for modulating the immune system, cell development, differentiation, and apoptosis (32, 33). Aryl hydrocarbon receptor nuclear translocator (ARNT), the dimerization partner of AhR, also plays a critical role for normal β-cell function (34). Expression of 9 of 15 genes/proteins comprising this pathway were decreased in 2-wk islets, including Cdk4, Cdk2, Chek2, and Rbl1 (Fig. 3C, Supplemental Table S9) (24).

Pathways regulating neuronal function.

All of the sympathetic, parasympathetic, and sensory nervous systems innervate pancreatic islets and are important for regulating islet hormone secretion and glucose metabolism (18). Two neuronal pathways, synaptogenesis signaling and dopamine receptor signaling, were altered in 2-wk islets. Thirty genes/proteins comprising these pathways were differentially expressed, including Cplx2, Rab3a, Cdh13, Syt5, and Maob (Fig. 3D, Supplemental Table S10) (24). Synaptogenesis signaling was predicted to be activated in 2-wk islet (z score = 2.13), indicating the important of neuronal function in regulating islet maturation. The dopamine signaling pathway was predicted to be inhibited in 2-wk islets. Dopamine, a neurotransmitter that plays a critical role in central nervous system and neurological disorders, is also synthesized in β cells and acts as an autoparacrine signal to regulate GSIS and glucose homeostasis (35).

Upstream regulators and master regulators critical for islet development and function

Upstream regulators.

Ingenuity pathway analysis can identify upstream regulators and master regulators that can explain changes in gene and protein expression and biological activities. It was not surprising that the top candidates of upstream regulators altered in 2-wk islets compared with fetal islets were regulators modulating cell cycle, transcription, and translation, such as Cdkn2a, Tp53, Rbl1, retinoblastoma, Cdkn1a, Myc, Rabl6, Erbb2, and E2f (Table 3). These changes are consistent with a significantly higher rate of islet cell replication in late gestation and the early neonatal period (36). Beyond regulators modulating cell replication, top activated upstream regulators in 2-wk islets included Gata1, Pax6, Bdnf, Nlrp3, Neurod1, Ppar-γ, Ppargc1a, Smarcb1, Nr5a2, transcription factor EB, Irgm1, Crebbp, Nkx2-2, and HNF1a (see Table 3). In contrast, the top upstream regulators that were predicted to be inhibited in 2-wk islets included Csf2, epidermal growth factor receptor (Egfr), Foxm1, Vegf, amphiregulin, Tal1, RE-1 silencing transcription factor (Rest), estrogen receptor-α (Esr1), Mitf, Nkx2-3, and Gata6 (see Table 3). Several well-recognized transcription factors critical for β-cell and islet development and function were identified as important upstream regulators, such as Pax6, Neurod1, Ppar-γ, Ppargc1a, Nkx2.2, Crebbp, Hnf1a, Mitf, Gata6, and Foxm1 (see Table 3).

Table 3.

Top upstream regulators

Upstream regulator	Molecule type	Activation z score	P	No. of target genes
CDKN2A	Transcription regulator	5.39	3.53E-15	46
TP53	Transcription regulator	4.15	3.45E-27	179
SMARCB1	Transcription regulator	3.75	1.96E-05	20
RBL1	Transcription regulator	3.69	7.77E-09	17
PAX6	Transcription regulator	3.51	4.25E-05	22
CDKN1A	Kinase	3.49	3.12E-22	53
RB1	Transcription regulator	3.27	3.62E-07	44
PPARGC1A	Transcription regulator	2.75	1.10E-02	24
NLRP3	Other	2.70	2.21E-04	12
GATA1	Transcription regulator	2.53	9.72E-15	45
NEUROD1	Transcription regulator	2.38	1.44E-02	6
CREBBP	Transcription regulator	2.27	5.13E-03	27
NKX2-2	Transcription regulator	2.24	9.27E-06	6
TFEB	Transcription regulator	2.23	2.55E-02	5
BDNF	Growth factor	2.12	1.58E-04	28
IRGM1	Other	2.12	4.34E-04	8
NR5A2	Ligand-dependent nuclear receptor	1.96	1.29E-02	10
PPARG	Ligand-dependent nuclear receptor	1.90	1.01E-05	41
HNF1A	Transcription regulator	1.73	2.36E-02	29
ESR1	Ligand-dependent nuclear receptor	–2.16	3.93E-06	91
NKX2-3	Transcription regulator	–2.28	3.40E-03	18
REST	Transcription regulator	–2.41	1.45E-03	15
GATA6	Transcription regulator	–2.44	4.91E-02	13
AREG	Growth factor	–2.72	7.70E-04	11
MITF	Transcription regulator	–3.04	2.42E-15	41
FOXM1	Transcription regulator	–3.10	1.41E-03	14
VEGF	Group	–3.28	5.08E-09	50
TAL1	Transcription regulator	–3.32	9.02E-03	17
E2F	Group	–3.65	7.88E-14	28
EGFR	Kinase	–3.81	5.88E-06	37
CSF2	Cytokine	–4.16	1.77E-06	45
RABL6	Other	–4.69	3.35E-15	22
ERBB2	Kinase	–4.87	2.76E-13	82
MYC	Transcription regulator	–6.81	1.23E-21	123

Interestingly, in contrast to Gata6 inhibition in 2-wk islets, Gata1 was predicted to be activated (z score = 2.53). Whereas Gata6 regulates development and differentiation of endoderm-derived tissues, including pancreas, Gata1 is critical for the development of the hematopoietic system, and regulates erythroid and megakaryocyte differentiation, hematopoietic stem cell proliferation, and the development of T lymphocytes (37, 38). Activation of Gata1 signaling suggests the importance of hematopoietic and immune systems in the development and normal function of islets and β cells. Both residential macrophages and T cells are identified in the pancreas, and they are critical for islet development and normal function (19, 20, 39). In fact, in our RNA-seq study, many immune-related and inflammation response pathways were identified among the top hits of pathways altered in 2-wk islets compared with fetal islets (Supplemental Table S11) (24). These pathways regulate differentiation and function of T cells, β cells, and macrophages, and are involved both in innate and adaptive immune systems. In addition to Gata1, several other upstream regulators critical for immune function and inflammation responses were also identified by IPA, such as Nlrp3, Irgm1, Gata6, Nkx2.3, Tal1, and CSF2 (Table 3).

Brain-derived neurotrophic factor (Bdnf) is a nerve growth factor and plays an important role in neuron and synapse growth, differentiation, and survival both in the central and peripheral nervous systems (40). Bdnf was predicted to be activated (z score = 2.12) in 2-wk islets and regulated the expression of 28 genes/proteins, including Cp, Cplx2, Gad2, Ins1, and Sst (Supplemental Table S12) (24). This suggests the importance of the nervous system in the maturation and function of islets. Consistent with the activation of Bdnf signaling, RE-1 silencing transcription factor (Rest), a master negative regulator of neurogenesis (41), was predicted inhibited (z score = –2.41) in the 2-wk islets, and regulated the expression of 15 genes/proteins in our data sets, such as Cdkn1b, Gad1, Hpgd, Nell1, and Scg2 (Supplemental Table S12) (24). Bdnf can regulate glucose metabolism and glucagon secretion, protect β cells, and restore insulin-secreting granules in diabetic mice (42, 43). Rest is expressed in the progenitors both of neurons and β cells during development, but is downregulated in differentiated cells (44). Repression of Rest is critical for β-cell development and function. Overexpression of Rest in β cells causes glucose intolerance, insulin exocytotic machinery disruption, and increased apoptosis (45, 46).

Nuclear receptor subfamily 5 group A member 2 (Nr5a2) is also known as liver receptor homolog-1. It is a transcription factor critical in regulating development, cholesterol transport, bile acid homeostasis, and steroidogenesis (47). It was predicted as an activated upstream regulator in 2-wk islets (z score = 1.96). Estrogen receptor-α (Esr1), another upstream regulator identified, can regulate the expression of Nr5a2. Nr5a2 and estrogen both can protect human islets against cytokine- and streptozotocin-induced apoptosis (48). Nr5a2 agonist can decrease immune-dependent inflammation of pancreas and increase β-cell mass and insulin secretion (49). Esr1 can maintain normal mitochondria and endoplasmic reticulum function and promote β-cell survival and insulin secretion (50).

Both Egfr and amphiregulin, a growth factor and agonist for EGFR, were predicted as inhibited upstream regulators in 2-wk islets. Egfr signaling promotes β-cell proliferation during pregnancy and in response to a high-fat diet (51). It also induces adult islet cell dedifferentiation (52). It was not surprising that Egfr signaling was inhibited in 2-wk islets, because β cells and other islet cells are well differentiated at this age and have much lower proliferation rates compared with fetal islets.

Transcription factor EB is a master regulator of lysosomal biogenesis and autophagy gene expression (53). It was an activated upstream regulator (z score = 2.23) in 2-wk islets, suggesting autophagy machinery was upregulated. Autophagy functions as a cell-protective mechanism, and is important for maintaining islet homeostasis and function (54). It also facilitates the compensatory increase in β-cell mass in response to a high-fat diet in mice. Glucagon-like peptide 1, whose RNA and protein levels both were increased in 2-wk islets, protects against β-cell death in animal models of type 2 diabetes partially by increasing lysosomal function and autophagic flux (55).

Master regulators.

The causal network analysis of IPA identifies potential novel master regulators responsible for the observed changes in gene or protein expression. The most relevant and interesting top master regulators altered in 2-wk islets are shown in Table 4. Dentin matrix acidic phosphoprotein 1 (Dmp1) is an extracellular matrix protein. It also plays an important role in regulating phosphate homeostasis (56). Depletion of phosphate significantly reduces the activity of phosphofructokinase-1 and impairs glucose-stimulated insulin secretion (57, 58). Dmp1 was predicted as an activated master regulator (z score = 6.91) and coordinated with 25 regulators to modulate expression of genes/proteins in 2-wk islets. Max dimerization protein 1 (Mxd1) is a member of the Myc network. It antagonizes Myc-mediated transcriptional activities by competing for the binding partner Max (59). Consistent with the inhibition of Myc in the 2-wk islets (see Table 3), Mxd1 was predicted as an activated master regulator with a z score of 5.44. Ubiquitin carboxyl-terminal esterase L1 (Uchl1) is a deubiquitinating enzyme and interacts with autophagy/lysosomal pathway and protects β cells from amyloid polypeptide–induced toxicity (60). Uchl1 was predicted as an activated master regulator in the 2-wk islets with a z score of 5.19. NADH:ubiquinone oxidoreductase subunit A13 (Ndufa13) is a subunit of the mitochondrial respiratory chain complex I. In addition to its electron transfer activity in mitochondria, Ndufa13 also interacts with signal transducer and activator of transcription 3 (STAT3) and acts as a negative regulator of STAT3 activity (61). STAT3 plays a role in β-cell neogenesis, and the activating STAT3 mutation is associated with impaired β-cell function (62). Histone deacetylase 7 is a transcriptional repressor. Histone deacetylase can modulate insulin signaling in β-cells through regulating the expression and activity of insulin receptor substrate 2 (63). Overexpression of histone deacetylase 7 impairs insulin secretion and is associated with type 2 diabetes (64). Birc5, also known as Survivin, is an inhibitor of apoptosis. Survivin is expressed in fetal and neonatal islets, and plays a role in pancreatic remodeling, islet homeostasis, and β-cell mass expansion (65, 66). Rabl6 is a member of the Ras superfamily of small GPTases. It regulates cell cycle and cell growth, and is a negative regulator of p53 and retinoblastoma tumor suppressors (67, 68). Rabl6 was predicted to be inhibited in 2-wk islets (z score = –5.52). MLX interacting protein like (Mlxipl), also known as carbohydrate-responsive element-binding protein, functions as a central metabolic coordinator. It is a glucose sensor and master regulator of lipogenesis (69, 70). Mlxipl also mediates glucose-stimulated β-cell proliferation in adult pancreas through regulating cell-cycle (71). It is inhibited in 2-wk islets with a z score of –4.85.

Table 4.

Top master regulators

Master regulator	Molecule type	Activation z score	P	No. of connected regulators
DMP1	Other	6.91	1.23E-22	25
MXD1	Transcription regulator	5.44	4.38E-27	52
UCHL1	Peptidase	5.19	1.34E-18	10
NDUFA13	Enzyme	4.70	1.48E-18	41
HDAC7	Transcription regulator	2.75	4.50E-22	48
TBP	Transcription regulator	–4.13	1.03E-18	75
MLXIPL	Transcription regulator	–4.85	1.95E-29	62
RABL6	Other	–5.52	4.53E-24	72
CDK4	Kinase	–5.80	2.23E-20	21
BIRC5	Other	–6.07	4.11E-30	17

Enriched proteins at either embryonic day 19 or 2 weeks

In addition to transcriptional regulation, protein expression and activity can be regulated by translation and/or posttranslational modifications. Our proteomics study revealed many proteins enriched either in e19 or 2-wk islets regardless of their transcript levels. There were 119 and 96 proteins enriched in e19 and 2-wk islets, respectively (Supplemental Table S13) (24). IPA of these 2 sets of enriched proteins indicated that the most enriched pathways in fetal islets were cell-cycle and transcription-regulating pathways. Sixteen cell-cycle and transcription regulating proteins enriched in fetal islets are listed in Supplemental Table S14 (24). These results were not surprising because proliferation of islet cells is still robust at this age during islet development. Four proteins, Rest, Brca1, protein jumonji, and nuclear receptor subfamily 2 group F member 6, all important for maintaining cell pluripotency, were enriched in fetal islets, suggesting β cells and other islet cells at this age were not completely differentiated nor functionally mature. This is consistent with dampened glucose responsiveness in fetal/neonatal β cells (13, 72).

The islet extracellular matrix modulates the microenvironment and cell-cell communication, and plays an important role in cell survival and function (73, 74). Three proteins regulating extracellular matrix were enriched in 2-wk islets, including collagen type VI alpha 5 chain, myosin heavy chain 11 (Myh11), and tumor necrosis factor receptor superfamily member 11b (Tnfrsf11b). Intact islet cell architecture and β-cell–β-cell interaction by intercellular junctions are important for glucose-stimulated insulin secretion (75, 76). Single β cells have decreased insulin biosynthesis and glucose-stimulated secretion. Four proteins regulating gap junction signaling were enriched in fetal islets, including gap junction protein α-1 (Gja1, also known as connexin 43), gap junction protein β-1 (Gjb1, also known as connexin 32), guanylate cyclase 1 soluble subunit α-1 (Gucy1a1), and natriuretic peptide receptor 3 (Npr3). Interestingly, young and adult β cells are enriched with connexin 36, instead of connexin 43 (77). Gap junctions are important for islet function and β-cell communication and synchronization. Functional maturation of β cells during development is concomitant with increased cell-cell coupling and gap junction density (76–78). Furthermore, 3 proteins regulating tight junction signaling were enriched in 2-wk islets, including NSF attachment protein β, Myh11, and Tnfrsf11b. Tight junction-associated proteins are increased during maturation of neonatal islets, and they may help to segregate the membrane portions enriched with hormone receptors and glucose transporters from those enriched with insulin-containing secretory granules (79, 80).

Discussion

The rodent islet functions as a mini-organ and contains 60% to 80% of β cells. In the present study, we sought to further elucidate cellular mechanisms modulating β-cell and islet development and functional maturation in intact islets by integrating the islet transcriptome and proteome. We hypothesized that changes in gene expression that correlated with changes in protein expression would identify key pathways that regulate β-cell and/or other islet cell development. To that end, we found expression changes of more than 900 genes that correlated surprisingly well with their protein levels. By comparing these overlapping genes/proteins between 2-wk and fetal islets, we confirmed previous findings that levels of key mitochondrial genes/proteins and NADH shuttle enzymes were all upregulated in 2-wk islets, indicating that glucose metabolic pathways and NADH shuttles may be more mature in 2-wk islets compared to fetal islets, as suggested by Jermendy et al in their rat study (13). As predicted, genes/proteins that regulate islet-cell proliferation, which is negatively correlated with maturity (14), were decreased in 2-wk islets. IPA showed that canonical pathways and upstream regulators modulating cell cycle, replication, and translation were among the top inhibited pathways and regulators in 2-wk islets. We also identified a number of pathways that have not been previously implicated in fetal islet and β-cell development, such as AMPK signaling, lipid and cholesterol metabolism, and AhR signaling.

Our finding that AMPK signaling was activated in 2-wk islets suggests AMPK signaling is associated with islet maturation. The activity of AMPK is regulated by glucose, amino acid, calcium, hormones, and cellular stress (81, 82). AMPK controls glucose sensing and uptake, lipid metabolism, glycogen, cholesterol and protein synthesis, and induction of mitochondrial biogenesis (83). The liver kinase B1–AMPK pathway regulates GSIS and β-cell mass dynamics (84-86). AMPK also plays a critical role in maintaining β-cell identity and responsiveness to glucose (25). Finally, deletion of both α1 and α2 catalytic subunit of AMPK in β cells strongly impairs insulin secretion in vivo (87, 88).

One of our more interesting observations was the differences in lipid and cholesterol homeostasis and signaling between 2-wk and fetal islets. Cholesterol is an essential component in cell membranes and lipid rafts and is required for the correct formation of secretory granules and insulin secretion in β cells. Fatty acid homeostasis and lipid signaling also regulate insulin secretion, β-cell survival, as well as β-cell proliferation (89–91). Cholesterol-deficient mice have a significantly decreased number of secretory granules as well as increased morphologically aberrant granules in the pancreas (89). Mice lacking cholesterol transporter Abca1 or overexpressed Srebp-2, a master regulator of cholesterol and fatty acid synthesis, in β cells have significantly elevated levels of total cholesterol and impaired GSIS (92, 93). Dysregulation of genes critical for cholesterol metabolism, such as LXR, low-density lipoprotein receptor, LDL receptor related protein 5, and low-density lipoprotein receptor related protein 6, impairs β-cell function and insulin secretion and is associated with the development of type 2 diabetes (94).

FXR/RXR plays an important role in regulating bile acid synthesis, lipid/cholesterol metabolism, and glucose homeostasis (95-97). It also plays a critical role in normal β-cell function, modulates GSIS, and protects islets from lipotoxicity (98). FXR/RXR regulates the expression of insulin, islet amyloid polypeptide (IAPP), somatostatin, MafA, Pdx1, Beta2/NeuroD1, and Adcy8 in β cells (98-100). Indeed, in addition to the changes of many genes/proteins for lipid metabolism, the gene/protein expression of insulin, IAPP, somatostatin, and gene expression of MafA and NeuroD1 were all increased in 2-wk islets, suggesting the association of FXR/RXR signaling and islet development and functional maturation. Thyroid hormone is critically important in islet development, maturation, and function (101, 102). Triiodothyronine activates Mafa transcription in islets and regulates β-cell maturation and glucose-stimulated insulin secretion (103). Thyrotropin (TSH) can stimulate the productions of thyroid hormone. It also directly regulates the expression of glucose transporter 2 and glucokinase in β cells and modulates glucose uptake and glucose-simulated insulin secretion (104). Our finding of increased expression of thyrotropin-releasing hormone (TRH), a TSH stimulator, in 2-wk islets suggests thyroid hormone and TSH signaling may play a role in regulating islet development and maturation.

Our finding that AhR signaling was significantly lower in 2-wk islets was intriguing. AhR and its dimerization partner ARNT both are important in regulating β-cell function and metabolism. AhR signaling also regulates the development and differentiation of the immune and nervous systems (105-107), both of which are key regulators of islet development and function. AhR-deficient mice have significantly reduced fasting insulin levels and insulin resistance, as well as decreased PPAR-α expression and altered circadian rhythm of genes regulating glucose and fatty acid metabolism (108, 109). ARNT modulates insulin secretion, glucose tolerance, and glucose sensing in β cells (34, 110). It also regulates the expression of important β-cell genes, including hepatocyte nuclear factor 4-α, glucose-6-phosphate isomerase, phosphofructokinase (PFK), insulin receptor, insulin receptor substrate 2, and Akt2 (34). Furthermore, through the common dimerization partner ARNT, AhR signaling can cross talk with hypoxia-inducible factor 1 pathway (111), another pathway regulating β-cell and islet function. Lower levels of key components of the AhR/ARNT signaling pathway at age 2 weeks compared to the fetus suggest that this pathway likely regulates β-cell and islet development rather than islet function (insulin secretion). This is consistent with the observation that long-term activation of AhR signaling is associated with type 2 diabetes (112).

Pancreatic innervation is critical for establishing islet architecture and spatial distribution during development, as well as cell-cell interaction and islet maturation and function (113, 114). Reduced islet innervation is associated with glucose intolerance and islet dysfunction in type 2 diabetes (115). Neuronal inputs fine-tune hormone secretion, regulate blood flow in islets, as well as regulate β-cell mass and proliferation (116, 117). Synaptogenesis signaling, dopamine signaling, and 2 upstream regulators Bdnf and Rest, which modulate neuronal function, were significantly different in 2-wk islets compared to fetal islets. Activation of synaptogenesis pathways in 2-wk islets suggests that this process continues after birth and likely plays a role in modulating secretion of insulin and other hormones from the islet. In contrast, the dopamine-signaling pathway was predicted to be inhibited in 2-wk compared to fetal islets. Dopamine inhibits insulin secretion, and lower levels of key genes/proteins in this pathway at 2-wk islets may contribute to the increase in GSIS that occurs after birth (118, 119).

IPA predicted many upstream regulators important for immune function that were altered in 2-wk islets. Although the changes in immune-related pathways were not observed in our analysis integrating the transcriptome and proteome data, many canonical pathways regulating immune system and inflammatory responses were altered in the transcriptome data set. This lack of correlation between RNA and protein expression may be due to the detection limit of proteomics, since many transcriptional factors and cytokines critical for modulating the immune system have low protein expression and/or short half-lives. Resident macrophages and T cells both are identified in the pancreas and are critical for islet development and normal function (19, 20, 39). AhR signaling, which was altered in 2-wk islets, also plays a critical role in regulating immune cell differentiation and function, as well as inflammatory responses (32, 120). Surprisingly, the transforming growth factor (TGF) β-1 protein, an important modulator of immune homeostasis and responses (121, 122), was detectable only in the fetal islets, suggesting that TGF β-1 protein may not be expressed in 2-wk islets or expressed only at a very low level. In addition to modulating immune function, TGF β-1 controls many cellular functions, including cell proliferation, differentiation, and apoptosis. TGF-β signaling also plays an important role in β-cell development, proliferation, and function. It promotes endocrine cell differentiation during organogenesis, and functions as a central regulator for β-cell proliferation (123, 124). Although inhibition of TGF-β signaling promotes β-cell replication (125), TGF-β can deregulate expression of epidermal growth factor (EGF), a β-cell proliferation promoter (126), and has synergistic effects in inducing cell proliferation (127, 128). In addition, inhibition of TGF-β signaling can lead to redifferentiation of human β cells expanded in vitro (129). However, it is unclear why TGF-β1 protein expression was decreased in the 2-wk islets. Further investigation in the cellular mechanisms controlling TGF-β1 expression and outcome of TGF-β1 downregulation during islet maturation process will provide more insights.

In conclusion, our present study integrated transcriptome and proteome data to elucidate cellular mechanisms controlling islet development and maturation in intact islets. In addition to confirming previously reported immature glucose metabolism and NADH shuttles in fetal islets, we also identified additional novel pathways that were associated with the maturation of islets such as cholesterol metabolism, islet innervation, AhR, and dopamine signaling. We also observed that key pathways regulating cell proliferation were already inhibited at age 2 weeks, corresponding to a marked increase in pathways regulating islet-cell differentiation and insulin secretion. Our study will serve as a valuable resource for further elucidating critical signaling pathways for β-cell and islet maturation and provides the basis for future research on diabetes prevention and pharmaceutical drug development, as well as islet transplantation.

Glossary

Abbreviations

2-wk

2-week-old

AhR

aryl hydrocarbon receptor

AMPK

adenosine monophosphate–activated protein kinase

ARNT

aryl hydrocarbon receptor nuclear translocator

Bdnf

brain-derived neurotrophic factor

Dmp1

dentin matrix acidic phosphoprotein 1

e19

embryonic day 19

Egfr

epidermal growth factor receptor

Esr1

estrogen receptor-α

FDR

false discovery rate

FXR

Farnesoid X receptor

GSIS

glucose-stimulated insulin secretion

IAPP

islet amyloid polypeptide

IPA

Ingenuity Pathway Analysis

LXR

liver X receptor

Mlxipl

MLX interacting protein like

Myh11

myosin heavy chain 11

Mxd1

Max dimerization protein 1

NADH

nicotinamide adenine dinucleotide

Ndufa13

NADH:ubiquinone oxidoreductase subunit A13

Nr5a2

nuclear receptor subfamily 5 group A member 2

PCA

principal component analysis

PPAR

peroxisome proliferator-activated receptor

Rest

RE-1 silencing transcription factor

RNA-seq

RNA sequencing

RXR

retinoid X receptor

STAT3

signal transducer and activator of transcription 3

TGF

transforming growth factor

Tnfrsf11b

tumor necrosis factor receptor superfamily member 11b

TR

thyroid hormone receptor

Uchl1

ubiquitin carboxyl-terminal esterase L1

Additional Information

Disclosure Summary: The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

Data Availability

All data generated or analyzed during this study are included in this published article or in the data repositories listed in “References.”